2013 International Conference on Advances in Social Science, Humanities, and Management (ASSHM 2013)

Research on characteristics in the spatial distribution of the whole county population of province, via ARCGIS analysis

Guiyao Zhou1, Yanyou Wu1*,Xianjian Xie2, Deke Xing1,Fang Fan3, Rui Yu1 1Key laboratory of Modern Agricultural Equipment and Technology, Ministry of Education＆Jiangsu province, Jiangsu University, Zhenjiang, China 212013;2Geography and Resource Science College , Normal University, Neijiang, China 641110; 3Software College, University, Chongqing, China 401331 [email protected], 1*[email protected],* Corresponding author Abstract: 1.Introduction 2000 and 2010 year census population data of Population,Resources,Environment, Energy each county of Sichuan Province, China has and Food problems are five major issues that the been chosen as research objects in our research. world is facing nowadays, with the population In this study, different spatial fitting model of problem being the core issue (CAS,2000).The geostatistics modules which are based on Arcgis spatial distribution of population is characterized Software has been used to discuss the fitting by the combination of results of geographical effects of population density data. The best conditions, climate suitability, degree of spatial distribution fitting model was obtained economy and social policy orientation and which has been used to make optimal Kriging others.The traditional sense of demographic data interpolation. The factors affecting Sichuan comes from national census every 10 years. It Province Population change have also been takes administrative areas such as counties and analyzed. Results showed that the overall cities as statistics unit.However, in practice there distribution of Sichuan province population has often exits some problems such as low spatial uneven spatial distribution and circular style. resolution data and administrative boundaries Counties population in space formed two not coinciding with population macroscopic distribution patterns: One is the boundaries,Regional population data can usually Western Sichuan-intensive area and the other only provide discrete and limited data sample. In one is basins gathering area. Time distribution of recent years, comparative studies on spatial different density region has changed over time, distribution of population density were carried especially the population of economic out by many authors (Feng et al.,2002; region and the Southern Sichuan economic Christopher;etal.,1999).Suggesting that people‘s region which are growing rapidly. The whole research methods on this subject of spatial population in Sichuan province shows an distribution of population data have gradually increasing trend from west to east. Natural changed from traditional statistics method to conditions, policies and economic conditions are diverse space data analysis method. The the most important factors that influence the emergence of various geostatistical analysis Spatial and Temporal distribution regularity of software such as Arcgis Series, GS +, etc has population of Sichuan Province. created a broad application of conditions for Keywords: Geostatistics; Sichuan Province; spatial distribution statistics of population. There Distribution of population; Spatial interpolation are various geostatistical models such as model Gaussian, exponential, spherical models etc.

© 2013. The authors - Published by Atlantis Press 1000

However, on processing data spatialization, the differences and economic disparities, the selection of best space model directly relate to economic spatial differences are still apparent. the quality of spatial distribution of statistical Differences between economic regions and data. In this paper, the geostatistics methods differences in city and outside coexist at the have been made use of and county has been used same time (Zhang and Ren,2011). as administrative units to analyze and discuss the 2.2 Data sources and data analysis fifth and sixth national census population density Research data include following parts: data of all 181 counties and cities in Sichuan (1)Sichuan digitized maps, which were digital Province. The best models were chosen to scan on Sichuan 1:100000 base map by Arcgis simulate the population denstiys‘spatial Software.(2)Administrative Region boundary distribution in Sichuan Province, and kriging data.(3)The total counties, cities population and interpolation methods have been used to administrative area data in Sichuan in year 2000 simulate the best trend surface of spatial and 2010.These data comes from 《 Sichuan distribution of population density. Finally, the Statistical Yearbook 2000, 2010》.Due to the dynamic characteristics of population accelerating urbanization, the original distribution of each region have been explored administrative division‘s settings do not suit the within two censuses period. pace of development.Therefore,the local 2.Materials and methods government made some adjustments which lead 2.1 Study area to changes in administrative boundaries of many Sichuan Province is situated in the south counties during this decade. For example, the old west of China,upstream of Yangtze River and three districts in have been surrounded by Chongqing,Yunnan,Tibet, changed into Daying County. In order to and other provinces.Its geographical facilitate comparison and analysis process, the coordinatesare97°21′—108°31′E,26°03′—34°19′ fifth census data has been used to do the N. Sichuan covers an area of 48.5×104km2, appropriate conversion, which is based on the ranking as the fifth largest Province in Western sixth census of the administrative unit divisions. China.It mainly constitutes of plateaus, In the first place, the 2000 and 2010 year mountains, hills and other topographies. Western population data was fed into the county's topography is high, while Eastern topography is population density database and the equation lower. The Study area is shown in Figure 1. It is which is Population density = total population dominated by a subtropical climate. In year 2000, / total area was used to calculate out 2000 and the population of Sichuan Province was nearly 2010 year population density of all the 181 87 million, accounting to 23.7% of Western counties and cities. China Population. Sichuan Province consists of Chengdu, , Ya'an, Wenchuan and other 181 counties and cities. It is one of the richest regions with important national resources, and also an important ecological barrier to the upstream of Yangtze River. The economy in this area is flourishing, often ranked first in . Sichuan has a provincial situation with "large population,weak economic foundation, and under development." Due to the differences Figure1. Study area in natural, historical economic conditions, policy

1001

2.3 Research methods on spatial distribution,variation characteristics of Geostatistics, also called the geostatistical the population density database of all 181 method, has been put forward by the famous Counties and cities of Sichuan province. French statistician, G. Matheron after doing a 3.1.1Trend analysis large number of theoretical studies on the basis With Arcgis software, Trend Analysis of the gradual formation of a new branch of Tools provide 3D perspective view that convert statistics. It is based on regionalized variable and sampling sites in study areas into the view that use of variograms to research natural treat the interested property value as height, and phenomenon that not only has randomness but allow users to analyze the whole trend of the also structural or spatial correlation and data sets of sampling point from a different dependence. Geostatistics is generally based on perspective (Tang,2006).This tool on a certain the semivariogram and kriging interpolation extent can reflect the spatial phenomenon methods to simulate and fit the spatial modes. change on spatial region. Each vertical bar in Geostatistical semivariogram mainly trend analysis graph represents a data point‘s include Circular,spherical,Gaussian ， value (Height) and position. These points are Exponential and other models. Arcgis (9.3vesion) projected onto an east-west and a north-south includes 11 kinds of models. Each model direction orthogonal surface, thus Arcgis primarily through nugget (CO), partial sill value software can make a best-fit line through the (C), sill (CO+C), and nugget coefficient to projection point, and use it to impersonate the describe the regular pattern of spatial existent trend on a specific direction. If the lines distribution data. Kriging interpolation method are straight, this shows no trend existence on the includes simple Kriging, ordinary Kriging, space(Tang,2006).Therefore, trend surface universal Kriging and other methods. In practical analysis module of the GIS software was used to work, ordinary Kriging is often used to explore analyze the trend of the year 2000 and 2010 the spatial differential features of statistical data. Sichuan county population density, respectively. 3. Geostatistical analysis Results are listed in figure2. According to the county-level administrative units, population database of Sichuan Province, which was previously established,a spatial analysis of the database was done that included 2000 and 2010 year population density data of 181 counties and cities of Sichuan province and simulated out the Spatial distribution trends characteristics. Finally, based on the parameters under different models, Figure 2 (a): 2000 year trend chart the best forecasting model was chosen. 3.1 Exploratory data analysis Purpose of exploratory data analysis is to find out the internal law of data-setting the interpolation parameters and choose theoretical variogram model rationally.The best spatial interpolation can be made using this analysis. The exploratory analysis tools of Arcgis software was used to make exploratory analysis Figure 2 (b): 2010 year trend chart

1002

As can be seen from Figure2 (a) and (b), both 2000 and 2010 year population densities of Sichuan Province were displayed as a gradually increasing straight line from west to east on east-west direction (X-axis direction of extension),indicating that the pattern of population distribution on East–west direction did not change, and eastern population density is still far greater than the western population density in this decade.On the north-south Figure 3 (a): 2000 year semivariogram direction (Y-axis direction of extension), the population density of Sichuan on the south to north direction is expressed as an inverted "U"-shaped curve.The pattern of population distribution that centralized from south, north to the middle direction did not change, and central Sichuan regions still have high population density areas. It can also be deduced from the figure that although the 2000 and 2010 on the east-west direction are approximately an inclined Figure 3 (b): 2010 year semivariogram line, the population density data on north-south From figure3 (a) and (b) ,it can be seen that direction are complex function curves. Therefore, in both two years ,most data are distributed in when the Arcgis software is used to carry out bottom left or left partial to middle region and kriging Interpolation, trend with linear or few points are distributed in top region where quadratic functions cannot be removed high population density intersect zones with low separately. population density.Therefore,it can be concluded 3.1.2 Semivariogram cloud analysis that the spatial distribution of two years points The Semivariogram cloud represent the conform to the law that closer point in space has semivariogram and covariance values of all more similarities, furthest has greater difference. dataset sample points,and express them as 3.2 Model select distance function between two From the above analysis,Circular ， points(Tang,2006).Its abscissa denote the Spherical,Gaussian,Exponential and Stable distance between geometric center of two models were initially identified to carry out administrative units, while ordinate denote the Kriging spatial interpolation. When this is done, logarithmic population density‘s variogram the quality of Model parameters directly value. Generally, closer point in space has more determines the quality degree of Fitting effect. similarities and, the furthest has greater The model parameters mainly include Nugget difference.Therefore,a semivariogram cloud (CO), Sill Value (C+C0), Partial Sill(C), analysis was made on 2000 and 2010 year (C0/(C0+C)), Average error (AE) and Root county population density, Sichuan. Results are Mean Square Standardized (RMSS). If the listed in figure3. nugget coefficient is less than 25%, it reflects that the system has a strong spatial correlation; and more than75%, reflects that the system has a weak spatial correlation. If RMSS value is near

1003

one and AE is near zero, the model fitting effect 250people/km2 is relatively sparse population is better. Therefore, the above five models and 6 regions; 250～500people/km2 is relatively dense parameters were used to make choice analysis on population areas;500 ～ 1000people/km2is best fitting effects model of 2000 and 2010 year densely populated areas; more than 1000 county population density of Sichuan Province, people/km2 is extremely dense population areas. respectively. Selection analysis process is listed The interpolation results are shown in figure4. in table 1. Table1: Population density variogram model selection

Theoretical CO CO+C CO/CO+C AE RMSS

model 2000 2010 2000 2010 2000 2010 2000 2010 2000 2010

Circular 0.00 0.00 7.04 6.95 0 0.00 3.08×10-4 3.09×10-4 3.35×10-3 3.53×10-3

Spherical 0.00 0.00 6.13 6.05 0 0.00 7.85×10-4 7.81×10-4 8.17×10-3 8.56×10-3

Gaussian 0.39 0.42 7.66 7.38 5.05×10-2 0.06 1.54×10-5 1.22×10-4 1.91×10-3 1.74×10-3

Exponential 0.00 0.00 4.70 4.62 0 0.00 4.90×10-3 5.01×10-3 2.79×10-2 2.99×10-2

Stable 0.31 0.34 8.42 8.09 3.64×10-2 0.04 2.82×10-5 6.48×10-5 7.08×10-4 7.85×10-5 From table1, it can be concluded that both nugget coefficient (CO/CO+C) of 2000 and 2010 year population density data that were logarithmically transformed were 0 under circular, spherical and exponential models. Relatively speaking, nugget coefficient under Gaussian model is the largest, so it could not be used as the best model. For average error test Figure4.(a): 2000 year interpolation map indicator, the average error of circular model has the minimum nugget coefficient of ‗0‘. For RMSS test indicator, the circular model was the largest. Therefore, the model is the best Interpolation model for both 2000 and 2010 year spatial distribution of population density of each county, Sichuan province. 3.3 Interpolation map create The 3.1, 3.2 analysis results showed that the Figure 4(b): 2010 year interpolation map circular model enable 2000 and 2010 year spatial Summarized from figure 4 (a) and (b), distribution of population density of each county Sichuan population density has the following of Sichuan Province to achieve the best effect of spatial distribution patterns: interpolation. In order to make the data more  The overall pattern of county population visually expressive, the population density of density spatial distribution has significant Sichuan Province was divided into 6 levels on difference. Uneven distribution of overall the basis of grading principle and standard of space and circle structure on East-West, population density and combined with special South-north direction exist at the same spatial distribution pattern of population history time. of this province, that is, less than 40 people/km2 ①Axial variation: From the population is extremely sparse population areas; 40～100 density trend analysis in Figure 2.1, it can be people/km2 is sparsely populated areas;100～ noted that both the population densities on

1004

East-West and South-North direction had showed that distribution of population density significant differences. The population density start to expand clearly. This phenomenon is tended to increase from west to east and mainly caused by rapid social development. Chengdu has the highest population density ②The overall distribution is of 500 ～ value. In the south-north direction, population 1000people/km2 population density with density concentrate from south and north to relatively little change, but has an increasing central direction, showing an inverted "U" shape trend. Compared with the year 2000, population curve feature. Chengdu- area has the of , , , Cities most population concentration regions. The have grown rapidly, maybe because of the moderate gathering area distributed outside implementation of "Chengdu-Chongqing Neijiang, city and other level of density Economic Zone" policy enacted by the Chinese zones with outward expansion having government. descending order. ③The areas with 250 ～ 500 people/km2 ②Circle style pattern: Sichuan Province population density expand slowly during this population formed a typical circle structure; time, mainly growth in , Mianyang, Chengdu area with high population density value Ya'an, etc., and city. showed outward diffusion characteristics ④The sparsely populated region with 100～ towards East and West direction. Western 250people/km2 showed relatively small change, Sichuan (except Panzhihua region) is an main focus being on Liangshan autonomous obviously low population dense region, and region, internally growing towards to Chengdu.

Sichuan basin (Chengdu to the east area) with 4. Spatiotemporal variation of population high population density. On the whole, it formed analysis gathering area and Western Sichuan sparse area District which were two 4.1Natural factors macro patterns of population distribution of Natural conditions are direct influencing geographical units. factors for population spatial distribution, and  Having compared the charts of 2000 and are also the most fundamental factors for the 2010 year spatial distribution of population formation and development of regional structure density, It is not difficult to find that there of Sichuan Province.Geological and is larger time variation characteristics of topographical characteristics have obvious counties population density in Sichuan influence on population spatial distribution in province. Sichuan Province.Macro-regional landscape ①Area of region with high population pattern laid the foundation for the development density (more than 1000people/km2) increased of Sichuan Province. Western Sichuan is significantly during the last decade. These dominated by plateau terrain, while the Central, regions were found in Chengdu surroundings Southern and Eastern Sichuan are mainly and small region in Southern Sichuan in the year dominated by basins and hilly topography. 2000. However, the high value region gradually Western Sichuan and Eastern Sichuan have large expanded outward with high speed in the year altitude differences with horizontal development 2010. Distribution area of Chengdu high-density such as low-lying basins and hill terrains. region increased significantly, and the Regional climate also influences the population Suining- regions population density spatial distribution. Eastern and Western grew faster during this decade. Southeast Sichuan have large climatic differences. In Sichuan (including Neijiang, Zigong) also recent years, a variety of meteorological and geological disasters occurred in Western Sichuan

1005

regions, especially landslides and mudslides. mechanical form over time. Western Sichuan Geological disasters frequently take place in regions are sparsely populated with poor natural Western Sichuan regions in the rainy season resources, weak economic foundation and since the 5.12 Wenchuan earthquakes and 4.20 characteristics. Despite the increased regional Ya'an earthquakes. These played important roles population density in this decade, growth rate is in modifying the spatial and temporal slow. distribution of population of Western Sichuan to 5. Discussion some extent. This article is only based on Arcgis 4.2Policy factors geostatistical analysis module discussing the best The evolution of the political situation and model selection problem of GIS spatial analysis the government's policy on the population applications, and the use of 2000 and 2010 year distribution is a significant impact factor, and counties population density in Sichuan Province also a "growth agent" for the development as an example to verify this analysis. Finally, the process of the population. In order to bring a corresponding spatial and temporal distribution change in the pattern of western regions, with the economic development of this province Chinese government implemented the "western during this decade was achieved. However, only development" macro strategy as from the year the relative optimal fitting model was used in 2000. Sichuan is one of the main beneficiaries of this paper to solve the problem. In order to make this policy. Since then, a large number of the spatial expression of ecology, economy, foreign-funded enterprises have invested and climate science and other fields with GIS established a large number of factories, geostatistical method more intuitive, a variety of economic development zones, etc. in eastern and statistical methods for in-depth discussion and southern Sichuan. This attracted a large number continuous improvement and exploration should of workers from outside the province, which be integrated. made the population of these regions grow in Reference mechanical form, and formed a high density of [1] Sustainable Development Research Group population groups. It changed the regularity of of Chinese Academy of the temporal and spatial distribution of the Sciences,(2000).Chinese Sustainable population in Sichuan province to a certain Development Strategy Report, Science extent. Press, . 4.3 Economic factors [2] Feng J,(2002).Research on Spatial Development of the population and distribution of population density and its economic development are often synchronized. Evolutionary model in Economic prosperity will promote the growth of Hangzhoucity,GeographicalResearch21(5): population. ―Chengdu circle‖ is the highest 635-646． population density areas in Sichuan Province, [3] Christophers,CohenJE,(1999).Continental Chengdu being the capital of Sichuan Province physiography, climate and the global and also the largest city in . It is distribution of human population.Science the center for financial,commercial, Press Beijing:24-136. administrative, cultural,educational, and medical, [4] Tang G A,(2006) .ARCGIS GIS spatial sports and insurance transactions. Its economic analysis experimental Course, Science development or urbanization attracted people Press, Beijing. from outside to come to this province, which resulted in regional population growth in

1006